Settlement Change in Byzantine : An Assessment of Finds from the General Survey of Central

William ANDERSON Victoria, Australia

ABSTRACT

Since 1986, extensive surveys have been conducted annually by Japanese Institute of Anatolian Archaeology. The surveys cover a large area of central , within the provinces of Kırşehir, , , and Aksaray, and to date, around a thousand sites have been recorded. Material collected from these sites forms a major resource of archaeological information for studying long-term settlement patterns in this region. In this study, sites dating from the Late Roman and Byzantine periods are assessed to investigate the nature and extent of settlement change at this time, and to consider how such trends inform understanding of the later phases of occupation at Kaman-Kalehöyük.

INTRODUCTION aim was first to clarify the chronology of sites, dividing those of the Early Byzantine (c. AD 330-630) and Surface artefact surveys are an important method of Middle Byzantine (c. AD 630-1100) periods, and second, assessing long-term settlement across wide regions, and to consider their location and character, to observe combined with historical and environmental data, surface consistencies and differences between the two periods. finds allow for specific phases of settlement change to be The main task involved examining assemblages that appraised. In central Anatolia, a significant shift in the had been identified in the published survey reports and pattern of settlement occurred from the seventh century site database as containing Byzantine material, as well AD, when Persian, and later, Arab armies invaded large as some which had been listed as Roman and Ottoman. areas of Byzantine territory in the east Mediterranean, Having determined the approximate date and character and advanced into Minor. Investigations of the of sites, their location, elevation and topography was seventh-century settlement shift have, until recently, assessed. focused on urban change, the transition from the Greco- Surveys have the advantage of being able to record Roman city of Late Antiquity to the fortified town of multi-period sites across large areas, are relatively the Byzantine era. However, archaeological surveys and inexpensive, and relatively non-destructive. However, environmental studies are increasingly yielding evidence there are a number of methodological limitations, for rural settlement trends, offering a more consistent not least the problem of identifying material, mostly view of the landscape as a whole. pottery, from widely-dispersed locations. Sites can This research approaches the question of Roman be approximately dated on the basis of diagnostic to Byzantine settlement shift through a preliminary potsherds – rims, bases and handles – but surface analysis of material collected on the general survey of pottery is often so fragmentary and degraded that central Anatolia (CAS). By examining these finds my identification through form is not always possible. 234 W. ANDERSON AAS XVII

Moreover, many ceramics from the late- and post-Roman period are locally produced, with no equivalents from stratified excavations. Added to this general problem of identification is the regionally-specific issue that few settlements of the Byzantine period have been excavated in central Turkey, the closest to the survey region being to the north, and and to the west.

EARLY BYZANTINE SITES

Early Byzantine sites are best identified by fine Fig.1 Red Slip Ware bowl from Çatal or Büyükteflek, Kırşehir ware pottery, primarily red-slipped wares (RSW) that province were mass-produced and widely circulated from the second to early seventh centuries AD (Hayes 1972). Production of RSW occurred at a number of coastal sites around the Mediterranean, but there were also some inland production centres, notably (Poblome 1999), and ‘imitations’ were made in several other parts of Asia Minor. The presence of RSW at a site indicates some level of Late Roman-Early Byzantine activity, but it is only when sherds from several different vessels are found, or in combination with other recognisably Byzantine material, that occupation can be confidently assigned. Being situated far from the coast and the major RSW production sites, this type of pottery is less common in central Anatolia than it is elsewhere, Fig.2 Rim of a dish with incised decoration from Eski however it was still present at several sites in the CAS Mandıra, Konya province region. As surface finds, RSW often survives only as tiny body sherds which are impossible to match with well- Mandıra (98-13) in Konya province (Fig.2). The top documented forms, and in just a few instances can sherds of the rim is decorated with incised wavy lines, which be given a more precise date than ‘Late Roman’. along with the rim’s form and fabric, are similar to One rare survival is an intact base and part body dishes excavated at (Pamukkale), identified as of a RSW bowl or dish, in the centre of which has been locally-produced fine wares dating from the sixth century scratched a cross, collected from Çatal, or Büyükteflek (Cottica 2000: 50, nos.17-19). (site 01-33) in Kırşehir province (Fig.1). The vessel is Coin finds are helpful for assessing site chronology made from fine, orange ‘Asia Minor fabric’ (5YR 8/8), in combination with retrieved pottery. Three coins were with few inclusions or voids, and has a fairly thick, among the assemblages examined. The earlier two, found darker red slip (2.5YR 5/8), with visible application at Gölcük (98-72) and Uçurdum (03-11) are small bronze marks. Its form, with slightly flared foot (diameter: issues of Constantius II, minted in , and dated 66mm), sunken belly, and thick body, resembles footed AD 351-355 (Kent 1981: 479, nos.96-98; 498, nos.104- bowls of the sixth century. Another distinctively Early 109). Both of these sites yielded RSW and other Late Byzantine sherd is the flat rim (diameter: 120mm) of a Roman pottery. The third coin, from Beşli Yeri (02- thin-walled vessel (fabric: 7.5YR 7/6), collected at Eski 158) near Kaman, is from the early seventh century, a 2008 Settlement Change in Byzantine Galatia 235 follis of and Heraclius Constantine, minted in with stabbing along them, a coiled handle, and a body , and dated AD 612-614 (Hahn 1981: 223, sherd with stabbing and incised wavy lines. Uçurdum is nos.159a, 159c, 160b). close to the Early Byzantine station of Andrapa that lay By far the most common material collected on on the Pilgrim’s Road between Ankara and Parnassos surveys is coarse ware pottery. With the exception of (Belke and Restle 1984: 126). Hacınınağıl (92-41), transport amphorae, which have been well studied and Azak Kalesi (99-28) and Sütce Höyük (00-04) in Konya classified, it can be difficult to accurately identify Late province, are all sites with typically Early Byzantine Roman coarse wares. Coarse wares can often be assumed cooking and storage vessel sherds of oxidised fabrics to have been locally produced, however they were also and including flat rims with stabbing, body sherds transported considerable distances. Other than observing with incised straight and wavy lines, and pithos sherds similarities in fabrics from localised sites, a fabric type with applied strips with thumbing, characteristics that series is unlikely to be achieved when considering such a resemble coarse wares found in Early Byzantine contexts large area. There are some features of pottery production at Tavium (Gerber 2003) and Pessinus (Devos 2003). that were more common in Early Byzantine times, for As is regularly the case, the Early Byzantine sherds instance, coarse wares of this period from the Acropolis were often alongside Roman and Hellenistic pottery, at Pessinus were found to have oxidised cores, in contrast suggesting long-term habitation throughout classical to the reduced fabrics of the preceding era (Thoen 2003: antiquity. 83). The function and character of sites is difficult to Recognisable forms of Byzantine coarse ware accurately gauge on the basis of a few surface sherds, include arch-rimmed cooking pots, flat-rimmed basins but there are often clues about a site’s status, from the and pithoi, flat handles, and external and internal composition of the assemblage and from the site itself. ribbing on the body, but these are not always enough to Large numbers of imported fine ware sherds are likely distinguish an assemblage as being Early Byzantine, as to signal a well-connected settlement with some level such features continued long into the medieval period. of prosperity, while assemblages consisting of mainly Decoration is one way in which coarse ware pottery can storage and transport vessels may be seen as places be at least designated Byzantine, although there was of agricultural activity. Some of the CAS sites have also considerable continuity in the surface treatment of architectural remains, such as Yapı (00-34) with its arch- cooking and storage vessels, both throughout and after shaped feature, Üçayak Kırşehir (01-18) where there are the Byzantine era. remains of a Middle Byzantine church with intact wall Early Byzantine coarse ware vessels can be broadly plaster, and Sütce Höyük (00-04), where the remains of divided between cooking and table wares, and thicker- city walls were observed. Although surface potsherds, walled pottery for storage and transport vessels, as even considered in their architectural context, only well as tiles and piping. The decoration on both thin allow for a basic level of site differentiation, the Early and thick-walled vessels features recurring techniques, Byzantine material suggests a diversity of site functions. particularly methods of incising the surface such as The predominence of storage vessels indicates however rouletting, combing, stabbing or gouging, and finger that the majority of sites had an agricultural purpose. impressing. Such decoration allows for pottery to be identified as Byzantine, but only when appearing in combination with fine wares or other chronological MIDDLE BYZANTINE SITES indicators can it be assigned to the Early or Middle period. While RSW pottery can indicate Early Byzantine The assemblage from Uçurdum (03-11) in Ankara settlement, Middle Byzantine and later occupation is best province, a flat site near Tuz Gölü, where the coin of detected from the presence of glazed ceramics. Glazed Constantine II was found, includes RSW sherds and fine wares became prevalent in Anatolia from the middle typically Late Antique coarse wares such as flat rims of the seventh century, when the red wares with red 236 W. ANDERSON AAS XVII slip of the Roman period were replaced by white wares polychrome sgraffito ware, turquoise glazed white ware coated with white slip and lead glaze. Excavations of sherds, and a sherd with light pink fabric (5YR 7/6) and Byzantine sites in the east Mediterranean, particularly thin, yellow glaze, that may be an example of the rare Saraçhane in (Hayes 1992), have allowed Glazed White Ware I, dated to the seventh and eighth for several types of glazed pottery to be dated quite century, and usually found in the . A accurately, and recent research has clarified regional sherd of Glazed White Ware I excavated from Kaman- chronology and production centres (Vroom 2003; Kalehöyük was identified by Joanita Vroom as being the Böhlendorf-Arslan 2004). first of its kind from a central or eastern part of Turkey Production of glazed ceramics continued long (Vroom 2006: 164-165), and this may be the second. after the end of Byzantine rule, and even into the Alongside these glazed fine wares, mainly from open modern era. Just as there is chronological uncertainty in vessels such as bowls and dishes, were several unglazed, distinguishing between Roman and Early Byzantine red cooking pots with combed decoration. wares, it is often hard to tell Byzantine from Seljuk and Another assemblage with a colourful range of Ottoman glazed ceramics: several sites in the CAS region glazed ceramics comes from Kışla (99-47), a hilltop site have glazed pottery, but much of this belongs to the post- in Konya province, close to Zengicek Kalesi (Belke and Byzantine period. Following the in Restle 1984: 245). Several sherds of turquoise glazed AD 1071, central Anatolia was under Seljuk control, but white ware, yellow and brown glazed pottery on pinkish that is not to say that Byzantine influence in the region fabrics with pale yellow slip, and painted, sgraffito wares ended altogether. Political change did not necessarily were collected here (Fig.4). The site also features coarse have a sudden impact on pottery producing traditions, wares in dark red fabric with typically Byzantine incised although the arrival of Seljuks and Ottomans in Anatolia decoration. The juxtaposition of these incised coarse would undoubtedly have affected cultural practices. wares with glazed fine wares illustrates the degree of Other external influences, for example the impact of continuity in coarse ware pottery production from the Normans and Franks from western Europe, should also Early to Middle Byzantine periods. be taken into account. Coarse wares are far more common at Middle Some of the CAS sites display clearly Middle Byzantine sites than fine wares, and the sites of Byzantine elements, such as Elemenlihöyüğü (03-28) Elemenlihöyüğü and Kışla are unusally rich. Many in (Fig.3). This assemblage includes medieval sites, especially those in remote locations, were brown and yellow glazed ware, a sherd of green glazed small, poor settlements, where little material culture pottery of white fabric, perhaps Glazed White Ware survives, often only undecorated body sherds that are II (dated ninth to twelfth centuries AD), two pieces of nearly impossible to date. For this reason, many sites

Fig.3 Surface finds from Elemenlihöyüğü, Ankara province Fig.4 Surface finds from Kışla, Konya province 2008 Settlement Change in Byzantine Galatia 237 that were probably occupied in the medieval period High levels of Late Roman to Byzantine settlement could not be identified as such, and it is only when in this region have already been studied as part of the other chronological indicators are present such as coins, Konya Plain Survey (Baird 2004) which identified 85 fine wares or architecture, that they can be designated Early Byzantine sites, an increase from the Roman Middle Byzantine. Little can be said about the function period. The foundation of new settlements is seen as and character of such sites, other than those with richer indicating population growth, a theory supported by an assemblages, which may be exceptional. Many Middle increase in aggregative site area. This is interpreted as Byzantine sites appear to have been short-lived, had resulting from agricultural intensification and greater fairly poor material culture, and featured less of the large investment in irrigation and cultivation, which could also storage and transport vessels evident from the earlier explain the fairly sudden downturn in settlement numbers period. and size from the seventh century. The position of sites – whether they were sited on mounds or flat land – was also considered. In discussing SITE LOCATION AND CHARACTER the preliminary survey findings, attention was drawn to the different locations of Roman and post-Roman sites Having inspected material from over 200 sites, it (Omura 2002: 61), and in more recent reports, it was was possible in most cases to assign them to either the suggested that the settlement transition from mounds Early or Middle Byzantine periods. Few sites featured to flat sites occurred in the Roman period (Omura pottery from both of these periods, and there was a 2006: 72). This would correspond with the tendency noticeable difference in the composition of the non- for Romans to occupy flat, lowland sites, while in Byzantine elements of the assemblages, with Early the later period, settlement shifted to more defensible Byzantine sites often featuring material from the Roman locations, as would seem to be the case with the medieval period, whereas Middle Byzantine and Seljuk-Ottoman re-occupation at Kaman-Kalehöyük. Although there was material often came from sites with evidence of Bronze not a great deal of variation between mound and flat and Iron Age habitation, but without Roman material. site occupation, there is a preference for flat sites in the For the purposes of this research, sites that probably did earlier period, with 64 per cent on flat sites, while for date from the Byzantine period, but which could not be the Middle Byzantine period the figure is 51 per cent. confidently identified as either Early or Middle, were These statistics should be qualified, however, because categorised as uncertain. the definition of a mound can be rather subjective and In total, 58 sites were identified as Early Byzantine, does not take into account a site’s broader topography. 35 as Middle Byzantine, and 49 were uncertain, while To obtain a more accurate impression of change in the the rest were either pre- or post-Byzantine. The lower occupation of mound and flat sites it is necessary to number of Middle Byzantine sites is perhaps explainable consider a longer timespan. by the difficulty in identifying those places which lack An important factor when considering site locations diagnostic sherds, but it is also widely acknowledged that is their elevation and topography. The average (mean) the number of settlements did fall during and after the elevation of all the Early Byzantine sites was 1033.60m seventh century. Having identified a sample of Early and above sea level, while the average for the Middle Middle Byzantine sites, it was then possible to examine Byzantine sites was 1061.29m. Although this is not a more closely the location and topography of these places. substantial difference, it could be partly explained by the The first aspect of the sites to be considered was variation in valley level across a large area, from around their geographical location. Despite the methodological 900 to 1200m. However, even when sites in a smaller inconsistencies of extensive survey, there were clearly area are considered, the difference remains minimal. areas with higher numbers of Byzantine sites, most Of the 37 Byzantine sites in Konya province, 22 dated notably a cluster in the southwest of the survey region. from the Early period, and 15 from the later period. Of the 93 Byzantine sites, 37 were in Konya province. The average elevation for the earlier sites is 1001.45m, 238 W. ANDERSON AAS XVII

period: few sites were continuously inhabited from Roman to medieval times. This reflects major social and economic upheaval, perhaps caused by military invasion, that affected patterns of agricultural production, trade and settlement structure. But the changes, at least in site location, are not as pronounced as one might expect, considering the conclusions of previous studies of Byzantine sites in the region (Belke and Restle 1984). A likely reason for this is the practical limitations of extensive survey, especially site selection, and the problem of pottery identification and site chronology, where sites cannot be dated accurately enough for settlements of the seventh to ninth centuries to be distinguished from those of the ninth to eleventh Fig.5 Chart showing the topographical location of Early and centuries. The real outcome of the seventh-century shift Middle Byzantine sites might therefore be overlooked, as the poorer, short-lived sites, which only feature a few sherds of undecorated and for the later period 1027.73, perhaps suggesting a pottery, cannot be confidently dated, and so most Middle preference for upland settlements in Middle , Byzantine sites are likely to date from the later, more but again, a small difference. More detailed investigation prosperous period. of site elevations in the Lake Tuz basin also found only The question of settlement shift in the Early moderate change in the location of post-Iron Age sites to Middle Byzantine period is particularly relevent (Kashima 2001). for the occupation history of Kaman-Kalehöyük, The modest change in site elevation over time which after centuries of continuous settlement, was might be explained by the region’s already high abandoned at some point in the , and position on the Anatolian plateau, but when looking at only re-occupied in Middle Byzantine times, a break site topography, the location of sites in relation to the of almost a thousand years. During the Late Roman surrounding landscape, a more pronounced difference period, settlement levels increased greatly across the emerges (Fig.5). Sites were divided into lowland (those entire landscape, with previously unoccupied areas of located on the valley plain), hillside and hilltop. The the countryside being settled, and growth in the number majority (84 per cent) of Early Byzantine sites were in and size of villages. So it is interesting to ask why there lowland positions, with 13 per cent on hillsides and only is little evidence of settlement at Kaman-Kalehöyük at one hilltop site, while of Middle Byzantine sites, 46 per this time. The tendency for long-inhabited höyüks to be cent were lowland, 37 per cent hillside, and 6 sites were abandoned in the Roman era but re-occupied in post- on hilltops. Also noteworthy is that most ‘uncertain’ Roman times can be seen at many sites in Anatolia. In the sites which had sparse material culture and no fine environs of Kaman-Kalehöyük there were several Early ware pottery to identify them by, were in mountainous and Middle Byzantine sites, agricultural, military and locations, suggesting that they date from the later period, ecclesiastical in nature. Further research should therefore and were temporary or short-lived settlements. concentrate on a smaller district, to clarify the processes and outcomes of settlement change within a distinct geographical and ecological zone. It should also address CONCLUSION how surface artefacts can inform us about the function and character of sites as well as their chronology. Material from the CAS survey indicates a definite settlement shift from the Early to Middle Byzantine 2008 Settlement Change in Byzantine Galatia 239

ACKNOWLEDGEMENTS Heraclius' bis Leo III, Alleinregierung (610- 720), Vienna. I am very grateful to Dr. Sachihiro Omura and the Hayes, J.W. team at Kaman-Kalehöyük for their kind hospitality 1972 Late Roman Pottery, London. and assistance with my research. This research was 1992 Excavations at Saraçhane in Istanbul, Volume 2, made possible thanks to a PhD. Fieldwork Grant from Princeton. Faculty of Arts, University of Melbourne, and a Spencer- Kashima, K. Pappas Trust Grant from the Centre for Classics and 2001 “The Relationship Between the Distribution of Archaeology, University of Melbourne. Archaeological Sites and Their Geomorphologic Conditions in the Lake Tuz Basin, Central Turkey,” AAS X, pp.129-139. BIBLIOGRAPHY Kent, J.P.C. 1981 The Roman Imperial Coinage, Volume VIII, The Baird, D. Family of Constantine I AD 337-364, London. 2004 “Settlement expansion on the Konya Plain, Omura, S. Anatolia: 5th-7th Centuries A.D.,” in W. 2002 “Preliminary Report of the General Survey in Bowden, L. Lavan and C. Machado (eds), Central Anatolia (2001),” AAS XI, pp.45-112. Recent Research on the Late Antique 2006 “Preliminary Report of the 2005 General Survey Countryside, Leiden, pp.219-246. in Central Anatolia,” AAS XV, pp.63-102. Belke, K. and M. Restle Poblome, J. 1984 Galatien und Lykaonien. Tabula Imperii 1999 Sagalassos Red Slip Ware. Typology and Byzantini 4, Vienna. Chronology, Turnhout. Böhlendorf-Arslan, B. Thoen, H. 2004 Glasierte byzantinische Keramik aus der Türkei, 2003 “Burial Goods,” in J. Devreker, H. Thoen and Istanbul. F. Vermeulen, Excavations in Pessinus: the Brown, L. so-called Acropolis. From Hellenistic and 2000 “The Pottery,” in R.M. Harrison, Amorium Roman cemetery to Byzantine castle, Ghent, Excavations 1989: the second preliminary pp.59-117. report, Anatolian Studies 40, pp.205-218. Vroom, J. Cottica, D. 2003 After Antiquity: Ceramics and Society in the 2000 “Late Roman Imported and Locally Produced Aegean from the 7th to the 20th Century A.C. Pottery from Hierapolis (Pamukkale, Turkey): A Case Study from Boeotia, Central Greece, Preliminary Evidence”, Rei Cretariae Romanae Leiden. Favtorvm Acta 36, pp.49-56. 2006 “Some Byzantine Pottery Finds from Kaman- Devos, G. Kalehöyük: A First Observation,” AAS XV, 2003 “The Pithoi,” in J. Devreker, H. Thoen and pp.163-169. F. Vermeulen, Excavations in Pessinus: the so-called Acropolis. From Hellenistic and Roman cemetery to Byzantine castle, Ghent, pp.369-374. William Anderson Gerber, C. Centre for Classics and Archaeology 2003 “Die Keramikgruppen von Tavium/ The University of Melbourne Büyüknefes,” Anatolia Antiqua 11, pp.223-251. Victoria 3010 Hahn, W. Australia 1981 Moneta Imperii Byzantini Band III, Von